The 59 and 39 domains of yeast U6 snrna: Lsm proteins facilitate binding of Prp24 protein to the U6 telestem region

Transcription

1 RNA (2002), 8: Cambridge University Press+ Printed in the USA+ Copyright 2002 RNA Society+ DOI: S The 59 and 39 domains of yeast U6 snrna: Lsm proteins facilitate binding of Prp24 protein to the U6 telestem region DANIEL E. RYAN, SCOTT W. STEVENS, and JOHN ABELSON Division of Biology , California Institute of Technology, Pasadena, California 91125, USA ABSTRACT The 59 and 39 domains of yeast U6 snrna contain sequences that are thought to be important for binding to Prp24 and Lsm proteins. By extensive mutational analysis of yeast U6 snrna, we confirmed that the 39 terminal uridine tract of U6 snrna is important for U6 binding to Lsm proteins in yeast. Binding of Prp24 protein to U6 RNA is dependent on or is strongly enhanced by U6 binding of Lsm proteins. This supports a model for U6 snrnp assembly in which U6 RNA binds to the Lsm2 8 core prior to binding Prp24 protein. Using compensatory base-pairing analysis, we show that at least half of the recently identified U6 telestem as well as a nucleotide sequence in the other half of the telestem are important for binding of U6 RNA to Prp24 protein. Surprisingly, disruption of base pairing in the unconfirmed half of the telestem enhanced U6 Prp24 binding. Truncation of the entire 39 terminal domain or nearly the entire 59 terminal domain of yeast U6 allowed for detectable levels of splicing to proceed in vitro. In addition to gaining knowledge of the function of the 59 and 39 domains of yeast U6, our results help define the minimal set of requirements for yeast U6 RNA function in splicing. We present a revised secondary structural model of yeast U6 snrna in free U6 snrnps. Keywords: Lsm4; pre-mrna; snrnps; spliceosome; splicing; U6 pseudoknot INTRODUCTION Splicing of pre-messenger RNA transcripts of eukaryotic genes is carried out in ribonucleoprotein particles called spliceosomes (reviewed in Nilsen, 1998; Burge et al+, 1999; Hastings & Krainer, 2001)+ These are composed of the pre-mrna substrate, the spliceosomal snrnas (U1, U2, U4, U5, and U6 snrnas), and about 70 known proteins, a number that is thought to be close to the actual total in a typical spliceosome (see Stevens et al+, 2002)+ A primary driving force for research in the field has been the overarching goal of understanding the specific roles of the spliceosomal snrnas and their associated proteins in catalyzing the pre-mrna splicing reaction+ The importance of U6 snrna in the splicing reaction is supported by several lines of evidence, some of which are briefly outlined here+ U6 snrna is the most highly conserved snrna component (Brow & Guthrie, 1988), and it is located at or near the active site of splicing catalysis during at least the first chemical step (Wassarman & Steitz, 1992; Kandels-Lewis & Reprint requests to: John Abelson, Division of Biology , California Institute of Technology, Pasadena, California 91125, USA; abelsonj@caltech+edu Séraphin, 1993; Lesser & Guthrie, 1993; Sontheimer & Steitz, 1993; Kim & Abelson, 1996; see also Valadkhan & Manley, 2001)+ Sequences in the 59 and 39 domains of yeast U6 snrna (nt 1 48 and ) are thought to be important for binding to Prp24 and Lsm proteins (Fig+ 1)+ Truncation of U6 RNA to remove nt from its 39 terminus hindered the normal binding of U6 RNA to Prp24 and Lsm proteins (Vidal et al+, 1999)+ Only one mutation of yeast U6 RNA has been identified as having an effect on its binding to these proteins+ The U6 point mutation, A91G, was found to inhibit the normal binding of Prp24 to U6 RNA in yeast extract (Shannon & Guthrie, 1991)+ We recently found that Prp24 and Lsm2 8 proteins are the only stably associated protein components of the yeast U6 snrnp (Stevens et al+, 2001)+ Recombinant Prp24 protein binds to gel-purified yeast U6 transcript with a dissociation constant (K d )of 100 nm, and the footprint of this protein bound to yeast U6 RNA has been mapped, primarily to the region of nt (Ghetti et al+, 1995)+ The normal binding of Prp24 protein to U6 is important for formation of U4-U6 di-snrnp during spliceosome assembly and recycling (Shannon & Guthrie, 1991; Jandrositz & Guthrie, 1995; Raghunathan & Guthrie, 1998a)+

2 1012 D.E. Ryan et al. FIGURE 1. Secondary structure of U6 snrna in yeast S. cerevisiae+ Mutations in the second and third RNA recognition motifs (i+e+, RRM2 and RRM3) of Prp24 protein suppressed the cold-sensitive (cs) growth defect of a U6 point mutation that hyperstabilizes the 39 stem-loop in U6 RNA, A62G (Vidaver et al+, 1999), suggesting that these two RRMs of Prp24 may bind to U6 RNA and play a role in unwinding the 39 stem-loop+ Similarly, mutations in RRM3 of Prp24 suppressed the cs growth defect of a base-pair disrupting point mutation, U4+G14C, in stem II of U4-U6 di-snrnp (Shannon & Guthrie, 1991), suggesting that Prp24 may play a role in U4-U6 snrna annealing+ Both of these cs mutations in U6 and U4 snrnas were suppressed by U6 point mutations that colocalized to two regions of U6 snrna (Fortner et al+, 1994)+ The sequences of these two regions of U6, that is, nt and 86 95, are mutually complementary and could form an RNA duplex called the U6 telestem (Vidaver et al+, 1999)+ Vidaver et al+ found support for the existence of the U6 telestem in their genetic studies of U6 sequences in yeast+ In the presence of the cs mutation U6 A62G, the suppressor (of cs) effect of a pair of U6 mutations that disrupt the U6 telestem was reverted by including a complementary pair of U6 mutations that can restore telestem base pairing+ In additional experiments, the same disruption mutations of the U6 telestem were found to exacerbate the temperature-sensitive (ts) growth defect of mutant prp24-r158s (with a point mutation in RRM2 of Prp24)+ However, by including mutations that can restore telestem base pairing, this exacerbation effect was reversed+ These experiments suggest that the U6 telestem exists in vivo and that it may interact with Prp24 protein+ Immunoprecipitation of Prp24 from yeast extracts carrying the cs mutation in U4 snrna, U4+G14C, showed an aberrantly high level of wild-type Prp24 protein bound to the mutant U4-U6 snrnp (Shannon & Guthrie, 1991)+ The combined results suggest that Prp24 binds to U6 RNA via its RRMs and promotes U4-U6 snrna annealing for spliceosome assembly+ A hypothetical mechanism is that the RRMs of Prp24, especially RRM2, may bind to the U6 telestem and subsequently release the telestem to allow initiation of U4-U6 snrnp assembly (Vidaver et al+, 1999)+ This release could destabilize base pairing in the U6 39 stemloop duplex that must unwind and base pair to U4 snrna to generate U4-U6 snrnp for assembly into spliceosomes+ Prp24 has been shown to interact via two-hybrid assays with Lsm2, Lsm5, Lsm6, and Lsm7 proteins (Fromont-Racine et al+, 2000)+ The Lsm ( like Sm ) proteins belong to a family that includes the evolutionarily and functionally related Sm proteins+ The Sm proteins form a doughnut-shaped heteroheptamer that binds to each of the spliceosomal snrnas except U6 snrna (for Sm core structures, see Kambach et al+, 1999; Collins et al+, 2001; Mura et al+, 2001; Törö et al+, 2001)+ U6 snrna is associated with Lsm2 8 proteins, and these are thought to form a doughnut-shaped heteroheptamer bound to U6 (Achsel et al+, 1999; Mayes et al+, 1999; Vidal et al+, 1999; Bouveret et al+, 2000)+ U6 binding to the Lsm core appears to stabilize U6 snrna against degradation (Mayes et al+, 1999)+ Lsm proteins are sufficient to promote the annealing of human U4 and U6 RNA transcripts in vitro, suggesting that these proteins may play a similar role in the annealing of U4 and U6 snrnps in vivo (Achsel et al+, 1999)+ For human U6 snrna, the 39 terminal 12 nt were found to be necessary and sufficient for binding to the Lsm proteins, and the tract of five uridines at the 39 end was shown to be an important determinant for this binding (Achsel et al+, 1999)+ In contrast, for yeast U6 snrna, the last 18 nt at the 39 terminus are necessary but not sufficient for U6 binding of the Lsm core (Vidal et al+, 1999)+ Vidal et al+ found that the 39 terminal domain of yeast U6 RNA is necessary not only for binding to Lsm proteins but also for U6 crosslinking to Prp24 and other U6-proximal splicing proteins+ The 39 terminal domain of yeast U6 (nt ) is mostly divergent, although it maintains the ability to form U2-U6 helix II and has a short uridine tract at its 39 end, as found in other organisms+ In this report of mutations in yeast U6 snrna that affect splicing (see also Ryan & Abelson, 2002), we introduced mutations or deletions in the 59 and/or 39 domains of yeast U6 RNA, especially in the 59 and 39 telestem sequences, in the U6 sequence of U2-U6 helix II and in the uridine tract at the 39 end of U6+ We found that specific sequences in the 59 and 39 domains of yeast U6 snrna are important for binding to Prp24 and/or Lsm proteins+ However, despite their importance for Prp24 and Lsm binding, most of these sequences were not found to be important for splicing in vitro+ It has been known that yeast U6 lacking nt from its 39 end is able to reconstitute 34% of fulllength U6 RNA splicing activity (Fabrizio et al+, 1989), despite the loss of one telestem base pair plus 8 of 11

3 59 and 39 domains of yeast U6 snrna 1013 base pairs of U2-U6 helix II and the entire 39 terminal uridine tract+ We have analyzed these specific sequences individually for their effects on assembly of U6-containing snrnps, on spliceosome assembly, and on splicing in vitro+ To further our knowledge of the chemical requirements for U6 function in splicing, we also defined the largest truncations of the 59 and 39 terminal domains of yeast U6 RNA that allow for detection of splicing in vitro+ RESULTS Truncation of the 59 and 39 terminal domains of yeast U6 snrna The splicing activities of a few 39 truncated yeast U6 RNAs have been previously reported+ A U6 transcript lacking 18 nt from the 39 end, U (truncated at the FokI restriction endonuclease site), was found to reconstitute 34% of full-length U6 splicing activity when added to U6-depleted yeast extract, whereas U showed no splicing activity in the in vitro assay (Fabrizio et al+, 1989)+ Studies in Saccharomyces cerevisiae ofa39truncated U6 gene, U6+1 90, on a centromere plasmid as the sole copy of the U6 gene were not viable (Bordonné & Guthrie, 1992)+ We decided to test further truncations of the active U RNA using the same in vitro splicing assay as Fabrizio et al+ (1989)+ All U6 RNAs discussed in this report were prepared synthetically by template-directed ligation of synthetic RNA oligonucleotides using T4 DNA ligase+ To test the splicing activity of full-length and 39-truncated U6 RNAs, splicing assays were conducted in yeast extracts using [a- 32 P]uridine-labeled actin pre-mrna as the splicing substrate+ In the assay, endogenous U6 snrna in the extract is depleted by oligo-directed RNase H digestion using a deoxynucleotide (d1) complementary to nt in yeast U6 RNA (Fabrizio et al+, 1989)+ Synthetic U6 RNAs are then added to reconstitute U6 snrnps in vitro, and these RNAs are also radiolabeled with 32 P to monitor their stability in the yeast extract+ We prepared the U truncated RNA studied previously as well as further truncations, U6+1 91, 1 88, 1 86, 1 84, 1 81, and In preliminary experiments, U6+1 94, 1 91, and 1 88 reconstituted 35 40% of full-length U6 splicing activity (Table 1, part A; data not shown) as reported previously for U (Fabrizio et al+, 1989), whereas the slightly shorter construct U reconstituted only 20% of the splicing activity of full-length U6 (shorter U6 RNAs were tested subsequently; see below)+ We observed that the full-length U6 RNAs were routinely degraded in yeast extracts to shorter RNAs equivalent in length to truncations at or near the 39 end of U2-U6 helix II+ Truncated RNAs U6+1 94, 1 91, 1 81, and 1 80 showed partial or extensive degradation by the loss of one to a few nucleotides (e+g+, U TABLE 1+ Splicing yields for truncated U6 RNAs in yeast extract+ Truncated U6 RNAs Relative splicing yields a Mole ratio of lariat:mrna Part A U % b 1:1+2 U % b 1:1+1 U % b 1+2:1 U % b 1:1+1 U6 wild-type (1 112) 100% b 1:3+3 Endogenous U6 c 250% b 1:4+1 Part B U % d n/a U % d n/a U thioP e 4% d :1 d U6 wild-type f 100% 1:2+6 d Endogenous U6 c 180% 1:6+0 Part C U f,g 23%, 31% d :1 d U6 wild-type f 100% :1 d a Splicing yields for reconstituted, synthetic U6 RNAs and endogenous wild-type U6 RNA are normalized relative to the yield for reconstituted, synthetic full-length U6 RNA (U6 wild-type) as a control+ b In these particular experiments, splicing yields are based solely on mrna products+ Normally, a splicing yield is based on first- and second-step splicing products+ However, the U6 truncations showed discrepancies in lariat degradation+ c Endogenous wild-type U6 snrna in yeast extract that was not U6-depleted or reconstituted with synthetic U6 RNA+ d For duplicate samples+ e The last two linkages at the 39 terminus are racemic phosphorothioates+ f The last four linkages at the 39 terminus are racemic phosphorothioates+ g The first seven linkages at the 59 terminus are racemic phosphorothioates+ n/a: not applicable+ was degraded completely to U6+1 91/90) whereas U6+1 88, 1 86, and 1 84 were extensively degraded to lengths matching the gel mobility of U (data not shown), suggesting that this latter group of U6 RNAs may be degraded at their 39 ends until double-stranded U4-U6 stem II impedes further exonucleolytic degradation (U4-U6 stem II base pairing includes U6 nt 80 at the end of the duplex)+ To block single-strand specific, exonucleolytic degradation of synthetic U6 RNAs in extract, we incorporated four phosphorothioate linkages between the last 5 nt at the 39 end of fulllength U6 (U6+thioP) and two such phosphorothioate linkages at the 39 end of U (U thioP) via chemical synthesis (thus each phosphorothioate is racemic)+ Because previous work had shown that an (R P )- phosphorothioate incorporated at the linkage in U6 blocked splicing completely (Fabrizio & Abelson, 1992), we decided to use the U and 1 80 RNAs without 39 end phosphorothioates, as these were significantly less degraded in extract than the slightly longer U6 RNAs, as mentioned above+ The 39 terminal phosphorothioate linkages substantially blocked extractdependent degradation of exogenously added U6 RNAs (Figs+ 2 and 3; data not shown), presumably by inhib-

4 1014 D.E. Ryan et al. FIGURE 2. Splicing of 32 P-labeled actin pre-mrna in U6-depleted yeast extract reconstituted with truncated and full-length U6 RNAs+ Yeast splicing extract was treated (A: lanes 2 13; B: lanes 2 5) or not treated (lane 1) with a DNA oligonucleotide (d1) complementary to U6 snrna+ After incubation at 34 8C to allow native RNase H digestion of endogenous U6 snrna, aliquots of the U6-depleted extract were reconstituted by adding synthetic, 32 P-labeled U6 RNA as indicated+ In synthetic U RNA, the first seven linkages at the 59 end and the last four such linkages at the 39 end were racemic phosphorothioates to block degradation via endogenous exonucleases+ Synthetic U6 RNA labeled with an contained four such phosphorothioate linkages at its 39 end; U6 RNA labeled as +thiop had two such linkages at its 39 end+ Splicing of 32 P-labeled actin pre-mrna substrate was assayed at 25 8C for 30 min+ This substrate has one intron and gives rise to exon1 and lariat-exon2 intermediates after the first chemical step of splicing and to excised lariat intron and spliced mrna products after the second chemical step (as indicated pictorially)+ Total nucleic acid for each sample was separated on a denaturing polyacrylamide gel+ In A, U6-depleted extract was assayed twice in lanes 2 and 3 for residual splicing activity, and the averaged background signal was used to correct the levels of splicing products assayed in lanes 6 9+ A replicate U6-depleted extract was assayed in lanes 4 and 5 and was used for lanes Thus, the assays for lanes 6 9 were repeated in lanes In B, samples were prepared as in A+ As for all of our presentations of individual gels, even those which were obviously cut and pieced together as in B, all of the samples for the gel were prepared concurrently and were separated on one single gel+ iting single-strand specific exonucleases in the extracts (Pandolfi et al+, 1999)+ We tested a variety of U6 mutants with and without phosphorothioate linkages at their 39 termini, and we found that the phosphorothioate modifications had no detectable effect on splicing activity in vitro (see below)+ Therefore, at least for full-length U6 RNAs, the splicing activity of U6 with 39 terminal phosphorothioates reflects equivalently the splicing activity of nonphosphorothioylated U6 RNA+ In experiments using the phosphorothioate-substituted U6 RNAs to block 39 end degradation, we assayed the splicing activity of 39-end-truncated yeast U6 RNAs (Fig+ 2A, Table 1, part B)+ We observed that full-length synthetic U6+thioP reconstituted 55% of wild type splicing activity in duplicate samples (Fig+ 2A, cf+ lane 1 and lanes 9 and 13) as is typical (Fabrizio et al+, 1989; our experiments)+ The truncation U thioP had the largest 39-end truncation that reconstituted detectable splic-

5 59 and 39 domains of yeast U6 snrna 1015 FIGURE 3. Splicing of 32 P-labeled actin pre-mrna in U6-depleted yeast extract reconstituted with mutant and wild-type U6 RNAs+ U6 mutations were introduced in the telestem sequences (nt and 86 95) and in the 39 terminal domain (nt ) of yeast U6 RNA+ Samples were prepared as in Figure 2+ In synthetic U6 RNAs marked with an, the last four linkages at the 39 ends were racemic phosphorothioates+ In A, three aliquots of the U6-depleted extract were assayed and used for background correction as in Figure 2+ The assays in A were performed three years before those in B using the same stock of extract stored at 80 8C+ ing activity, and it produced a 4% splicing yield in duplicate samples (Fig+ 2A, lanes 8 and 12)+ In contrast, further truncations of the U6 RNA, that is, U and 1 80, effectively blocked splicing activity as no splicing products were detected (Fig+ 2A, lanes 6, 7, 10, and 11)+ In the original study of the splicing activity of U in vitro, it was noted that the ratio of lariat intron versus spliced mrna was times higher than for most U6 RNAs (Fabrizio et al+, 1989)+ We were able to reproduce this result exactly, and essentially the same results were obtained for U6+1 91, 1 88, 1 86 and 1 85+thioP (data not shown)+ For each of these truncated U6 RNAs, quantitation of the splicing products (on a gel exposed to a PhosphorImager screen) including the unusually abundant lariat intron, showed that the mole ratio of lariat intron versus mrna was very close to 1:1 in each case (ranging from 1+3:1 to1:1+2; Table 1)+ In contrast, the mole ratio of lariat intron versus mrna for full-length U6 reconstitutions was generally close to 1:3 in these experiments (ranging from 1:2+6to1:3+3)+ As the lariat intron and spliced message must be nascently generated in equimolar amounts, the lariat is typically degraded about threefold more than the spliced message in our wild-type extracts reconstituted with full-length U6+ Surprisingly, the truncated U6 RNAs apparently blocked the normal degradation of lariat intron+ In previous work to define the minimal sequence requirements for U6 function, human U6 snrna was successively truncated at its 59 end, and it was found that deletion of 23 nt from the 59 terminus had no deleterious effects on splicing in HeLa extract (Wolff & Bindereif, 1992)+ Up to 37 nt were deleted from the 59 terminus of human U6 RNA, and it still maintained detectable splicing activity+ However, this maximal truncation

6 1016 D.E. Ryan et al. showed only a,10% yield of splicing products+ A similarly large truncation had not been tested in yeast U6+ Therefore, we constructed a synthetic U6 RNA lacking the first 38 nt of the 59 end of yeast U6 and containing seven phosphorothioate linkages between the first 8 nt at the truncated 59 end and four such linkages at the 39 end to block exonucleolytic degradation in yeast extracts+ The U truncated RNA reconstituted 23 31% of full-length U6 splicing activity (Fig+ 2B, lane 4; Table 1, part C)+ Therefore, as reported for human U6 RNA, yeast U6 does not require the 59 terminal domain for detectable splicing activity+ Extensive mutation of the 39 domain and telestem sequences of yeast U6 snrna We have shown that truncation of the entire U2-U6 helix II region and 39 tail of U6 in U reconstituted about 30% of full-length U6 splicing activity and inhibited the normal degradation of intron lariat+ Truncation of the entire 39 domain (nt ), which removes the entire 39 sequence of the putative U6 telestem (nt 86 95), reduced splicing activity to 4% of full-length U6+ To assess the loss of sequence information separately from any effects of shortening the U6 RNA, we prepared U6 RNAs with multiple cytidine mutations in the 39 terminal domain, as there are only two cytidines among the 27 nt in the 39 domain of wild-type yeast U6+ In preliminary studies, we found that polycytidine (polyc) mutations in the U2-U6 helix II region of U6 led to extensive degradation of these RNAs in yeast extract, as we had observed for truncated U6 RNAs (see above)+ By synthetically incorporating four a-phosphorothioate nucleotides at the 39 termini of the U6 RNAs with helix II mutations, degradation by single-strand specific, exonuclease activity in the extract was almost completely blocked (theoretically 94% blocked based on racemic phosphorothioates)+ The polycytidine mutations studied in our experiments included a complete polycytidine substitution of the U6 sequence of U2-U6 helix II in mutant U polyC+thioP (+thiop means that phosphorothioates were incorporated at the 39 end)+ In our initial studies, the telestem mutations were U polyC, U polyC and U polyC and the two combinations of these upstream and downstream telestem mutations+ (Nucleotides of the telestem sequence were contained on another piece of U6 in our original ligation scheme and were not mutated in the initial studies+) The combined telestem and helix II mutations were U polyC+thioP and U polyC+thioP+ The polyC mutation is a complete conversion of the entire 39 terminal domain to polycytidine, except for maintenance of the four 39 terminal uridines+ As mentioned, the terminal uridine tract is important for binding of U6 RNA to Lsm proteins in human cells, and it was therefore implicated for U6 binding to Lsm proteins in yeast+ We investigated mutations of the 39 terminal uridine tract separately (see below)+ The effects of these polyc mutations of helix II, the telestem, and the 39 terminal domain on splicing in vitro are presented in Figure 3A and Table 2, Column A+ Extensive mutation of the 59 sequence of the telestem to polycytidine or complete mutation of the 39 telestem sequence to polycytidine resulted in a splicing yield that was 65 85% that of wild-type U6 (Fig+ 3A, lanes 5 7)+ Nearly complete mutation of the entire telestem (for U polyC,86 95polyC) or of the entire 39 terminal domain (for U polyC+thioP, etc+) resulted in a relative splicing yield of 35 55% (Fig+ 3A, lanes 8 12)+ The effects of mutating both the upstream and downstream sequences of the telestem were somewhat more deleterious than expected from the effects of each sequence mutated separately+ In Figure 3A, mutating the two sequences of the telestem simultaneously was about 1+5 times more deleterious than expected from the separate mutations+ To investigate complete disruption of the U6 telestem, we mutated the entire upstream or downstream telestem sequence to polycytidine in the mutants U polyC and U polyC+ To disrupt every G-C base pair in the telestem, we also tested these two polyc mutations in combination with a cytidine mutation of the only remaining guanine within the telestem sequences+ These two mutants, U polyC,86C and U6+39C,86 95polyC, have no possibility of a Watson Crick base pairing within the telestem sequences+ In our preliminary experiments, when the U6 reconstitutions were successful but not optimized, it was clear that there was no discernible difference between U polyC and U polyC,86C in replicate splicing assays; however, there was a small but reproducible difference in splicing activity between the U polyC and U6+39C,86 95polyC mutants+ The results for these telestem disruption mutations are presented in Figure 3B and Table 2, Column B+ The combined results (Table 2, Columns A and B) show that complete disruption of the telestem resulted in a fold reduction in splicing activity (Fig+ 3B, lanes 5 7, 10 12)+ We tested the mutations in Table 2, Column B with and without phosphorothioate linkages at their 39 termini and found that these modifications had no detectable effect on splicing activity+ For comparison to our initial data (Table 2, Column A), we also assayed the in vitro splicing activity of U polyC+ Splicing activity for this particular polyc mutant, regardless of whether stabilized against degradation in extract, was the most variable, ranging from 13 to 85% in our assays+ Similarly, our polycytidine mutations that overlap the U polyC region, that is, the U polyC and U polyC,86C mutations, produced a relatively broad range of splicing yields (14 48%, including data not

7 59 and 39 domains of yeast U6 snrna 1017 TABLE 2+ Splicing yields for mutations in the 39 terminal domain and telestem sequences of U6 RNA in yeast extract+ Relative splicing yields a Mutated U6 RNAs Column A Column B b Growth in yeast c U polyC 88%, 55% d n+t+ U polyC 85%, 60% e, 42% d 16%, 13% f lethal U polyC 48%, 45% f U polyC 68%, 41% e, 64% d 67%, 69% f U6+39C,86 95polyC 53%, 52% f n+t+ U polyC,86 95polyC 42% n+t+ U polyC,86 95polyC 35%, 35% e n+t+ U polyC,86 95polyC lethal U polyC g 50%, 52% e / U polyC g 55% / U polyC g 36%, 43% e U6 wild type 100% 100% a Splicing yields for reconstituted, synthetic U6 RNAs are normalized relative to the yield for reconstituted, synthetic wild-type U6 RNA as a control+ b The experiment for column B is shown in Figure 3B+ c Growth was assayed at 16, 23, 30 and 37 8C; : wild-type growth at all temperatures, / : a subpopulation of slow growing colonies appeared reproducibly at all temperatures except 37 8C, lethal: no growth at all temperatures, n+t+: not tested+ d Yield in a replicate experiment using the same extract three years later+ e Yield in a duplicate experiment+ f For the second yield listed, the last four linkages at the 39 terminus are racemic phosphorothioates+ g The last four linkages at the 39 terminus are racemic phosphorothioates+ shown)+ The range of splicing yields for freshly ligated U polyC (Table 2, Columns A and B) is more typical of the range we observe in replicate trials+ The unusually wide range observed for U polyC may be due to variable misfolding of the mutant RNA+ We note that in the more recent experiments (Table 2, Column B), the splicing yields determined for U polyC and the other U6 telestem disruption mutations appear to correlate best with the growth phenotypes we observed for these mutations in yeast cells (Table 2; see below)+ To determine whether these U6 telestem and 39 domain mutations are viable in yeast, we transformed a haploid strain, HM1 (Madhani et al+, 1990), containing a chromosomal deletion of the U6 gene (SNR6) that also contains a plasmid copy of the U6 gene plus counterselectable URA3 marker gene+ Mutant U6 plasmids were constructed as described in Materials and Methods and transformed into yeast strain HM1, and the wild-type U6 gene was eliminated by selective growth on media containing 5-fluoroorotic acid (5-FOA; Boeke et al+, 1987)+ The U6 snrna mutations tested in vivo and the resulting growth phenotypes are presented in Table 2+ Upon counterselection of the wild-type U6 gene in strains containing U polyC or U polyC plasmids, we observed a substantially reduced colony number with respect to wild-type controls and other viable U6 mutations, and the colonies that did arise grew much more slowly+ To determine whether these colonies resulted from revertant or compensatory mutations in the mutant U6 genes, plasmids harvested from six colonies of each of the two slowgrowing strains were transformed into Escherichia coli, and the recovered U6 genes were sequenced for the 12 plasmids+ We found no revertant or compensatory mutations within the U polyC and U polyC genes+ Therefore, the slow-growing colonies are viable (1) perhaps without any additional mutations or changes in transcription levels, (2) as a result of spontaneous, extragenic suppressor mutations that were not identified, or (3) as a result of changes in the level of the mutant U6 RNA produced+ Regardless, our results demonstrate that neither the U6 sequence of U2-U6 helix II nor the U6 telestem is required for splicing activity or for growth at various temperatures in yeast+ For most studies of snrna mutations, the mutational effects found in yeast cells have correlated well with the effects observed in yeast extracts and vice versa (Fabrizio & Abelson, 1990; Madhani et al+, 1990; Madhani & Guthrie, 1992, 1994; McPheeters & Abelson, 1992; McPheeters, 1996)+ The viability of U6 telestem mutations in yeast correlates well with the modest effects of the telestem disruption mutations on premrna splicing in vitro (Table 2)+ Both yeast and mammalian U6 snrnas have a short uridine tract at their 39 ends that is thought to be an essential determinant for binding to Lsm proteins (Achsel et al+, 1999; Vidal et al+, 1999)+ We designed an experiment to test the splicing effects of mutating the four terminal uridines in yeast U6 RNA+ In two mutants,

8 1018 D.E. Ryan et al. these four uridines were entirely mutated to adenines or cytidines+ Conversely, to eliminate all other sequence information in the 39 terminal domain, we also tested our U6 mutant that possesses polycytidine mutation of the entire 39 domain except for the four terminal uridines+ In two additional variations, the terminal uridines were also mutated, either to polyadenine or polycytidine+ To stabilize the 39 terminus against exonucleolytic degradation, we incorporated four phosphorothioate linkages between the last 5 nt at the 39 end of each of these U6 RNAs+ Splicing assays for the 39 terminal mutants are presented in Figures 4A and 5A and Table 3+ Mutation of the entire 39 terminal domain except for the last four uridines (i+e+, nt ) resulted in two- to threefold less splicing activity than for wild-type U6 (Fig+ 3A, lane 10; Fig+ 4A, lane 14; Table 2, part A; Table 3), but further mutation to include the entire 39 terminal uridine tract caused splicing activity to drop to barely detectable levels (2% yields, Table 3; Fig+ 4A, lanes 12 and 13)+ Surprisingly, mutation of only the 39 terminal uridine tract to polyadenine or polycytidine had a rather minor effect on splicing in vitro (Fig+ 5A, lanes 3 and 4; Table 3)+ To ensure that a polyuridine tract was not added to the 39 ends of the U6 uridine tract mutants by some activity in the extract, we conducted a simple experiment to check whether the U6 uridine tract mutants could be uridinylated in extract+ Our standard depletion of endogenous U6 snrna in 10 ml of epitope-tagged Prp24(HA) 3 extract in splicing buffer was followed by addition of 75 mci of [a- 32 P]UTP and then addition of 25 fmol of wild-type U6 or 39 uridine tract mutant U6 RNA (U polyA/polyC)+ After incubation at 23 8C for 20 min, reconstituted U6 snrnps in these 25-mL FIGURE 4. A: Splicing of 32 P-labeled actin pre-mrna in U6-depleted yeast extract reconstituted with mutant and wild-type U6 RNAs+ U6 mutations were introduced in the telestem sequences (nt and 86 95) of yeast U6 RNA, including potentially hyperstabilizing mutations (lanes 9 and 10)+ Mutations were also introduced in the yeast U6 39 terminal domain (nt ) and terminal uridine tract (nt )+ Samples were prepared as in Figure 2+ Three aliquots of the U6-depleted extract were assayed and used for background correction as in Figure 2+ In synthetic U6 RNAs marked with an, the last four linkages at the 39 ends were racemic phosphorothioates+ B: U4-U6 U5 snrnp assembly for U6-depleted yeast extract reconstituted with mutant and wild-type 32 P-labeled U6 RNAs+ U6 mutations were introduced in the telestem sequences (nt and 86 95)+ Yeast splicing extract was treated with d1 oligonucleotide to digest endogenous U6 snrna as described in Figure 2+ Aliquots of the U6-depleted extract were reconstituted by addition of 2 fmol of synthetic, 32 P-labeled U6 RNA as indicated+ For lanes 3 6, samples were incubated at 23 8C for 20 min and then loaded onto a nondenaturing 4% polyacrylamide (79:1) gel for separation of the U6 snrnp-containing complexes as well as free U6 RNAs+ For lane 2, actin pre-mrna was added just prior to the 20-min incubation+

9 59 and 39 domains of yeast U6 snrna 1019 FIGURE 5. A: Splicing of 32 P-labeled actin pre-mrna in U6-depleted yeast extract reconstituted with mutant and wild-type U6 RNAs+ U6 mutations were introduced in the 39 terminal uridine tract (nt ) of yeast U6 RNA+ Samples were prepared as in Figure 2+ All samples were prepared concurrently and were separated on the same gel+ B: U4-U6 U5 snrnp assembly for U6-depleted yeast extract reconstituted with mutant and wild-type 32 P-labeled U6 RNAs+ U6 mutations were introduced in the 39 terminal uridine tract (nt ) and the 39 terminal domain (nt ) of yeast U6 RNA+ Samples were prepared as in Figure 4B+ In synthetic U6 RNAs marked with an, the last four linkages at the 39 ends were racemic phosphorothioates+ samples were immunoprecipitated with 0+5 mg of 12CA5 antibody on protein A-Sepharose beads (a background control lacked antibody)+ Scintillation counting of the washed beads showed that neither wild-type U6 nor the 39 uridine tract mutants were uridinylated by [a- 32 P]UTP under splicing conditions (data not shown)+ Therefore, no alternative 39 terminal uridine tract was added to our mutant U6 RNAs by factors in the yeast extract+ Depletion of Prp24 protein in yeast extract affects the assembly of U6 RNA into U4-U6 U5 tri-snrnps Raghunathan and Guthrie (1998a) have shown that Prp24 catalyzes the reassembly of U4-U6 snrnp during spliceosome recycling for successive rounds of splicing in yeast extracts+ They could not readily determine whether Prp24 is important for the biogenesis of U6- containing snrnps, as these complexes were already present in their Prp24-depleted extract+ To determine whether Prp24 is important for the production of U6- containing snrnps in our extracts, we employed our U6 reconstitution method to assay for assembly of 32 P- labeled U6 RNA into U6 snrnp complexes+ The standard protocol for RNase H depletion of endogenous U6 RNA is very effective at digesting the total endogenous U6 RNA in yeast extracts, including that in U4-U6- containing snrnps (Fabrizio et al+, 1989)+ We prepared extracts of a yeast strain (PRY112) carrying an epitope-tagged Prp24 gene as the sole copy of this essential gene+ This strain was originally prepared and used by P+ Raghunathan, and we received an aliquot of her extract for comparison with ours+ For all samples in our experiments, we first depleted the endogenous U6 RNA using our standard protocol+ Split samples were then either immunodepleted or mock depleted of epitope-tagged Prp24(HA) 3 protein using commercial

10 1020 D.E. Ryan et al. TABLE 3+ Data summary for mutations in the 39 terminal domain, 39 stem-loop and telestem sequences of yeast U6 RNA+ Mutated U6 RNAs a Splicing yields b U4/U6 U5 yield c Polyoma-Lsm4 I+P+ d Prp24(HA) 3 I+P+ e U CpolyC,86C 28% f (14 45%) 18% f,g (15 23%) 39%, 33% h 2%, 2% h U6+39C,86 95polyC 44% f (35 53%) 62% f,g (50 76%) 63%, 56% h 4%, 3% h U polyC 2% i 7% j 2% 8% U polyC, polyA 2% i 6% j 2% 6% U polyC 34% i (34 52%) 20% j 24% 8% U polyC 102%, 114% h 105% f ( %) 42% 140% f (83 228%) U polyA 80%, 83% h 64% k 10% 7%, 9% h U polyC 77%, 81% h 76% k 9% 15%, 9% h U6+U80G 5% 23%, 23% h,l 280% 940%, 900% h U6 wild type 100% 100% 100% 100% a The last four linkages at the 39 terminus are racemic phosphorothioates, except for U6+U80G, which has no phosphorothioates+ b Splicing yields for reconstituted, synthetic U6 RNAs are normalized relative to the yield for reconstituted, synthetic wild-type U6 RNA as a control+ c Yield of U4/U6-U5 tri-snrnp assembled from reconstituted U6 RNA, separated by native gel electrophoresis and normalized relative to the U6 wild-type control+ d Relative amount of reconstituted U6 RNA coimmunoprecipitated on anti-polyoma protein G-Sepharose beads, normalized relative to the U6 wild-type control+ e Relative amount of reconstituted U6 RNA coimmunoprecipitated with 12CA5 antibodies on protein A-Sepharose beads, normalized relative to the U6 wild-type control+ f Average of four or more trials (range of yields)+ g U4/U6-U5 tri-snrnp assembly gel showed aberrantly high levels of naked mutant U6 RNA not incorporated into U6 snrnps and a minor accumulation of free U6 snrnp relative to the U6 wild-type control+ h Yield in a duplicate experiment+ i Yield from experiment in Figure 4A (range of yields including additional trials+ j U4/U6-U5 tri-snrnp assembly gel showed aberrantly high levels of naked mutant U6 RNA not incorporated into U6 snrnps+ k U4/U6-U5 tri-snrnp assembly gel showed a minor accumulation of naked mutant U6 RNA relative to the U6 wild-type control+ l U4/U6-U5 tri-snrnp assembly gel showed aberrantly high levels of free U6 snrnp relative to the U6 wild-type control+ 12CA5 antibodies (as in Raghunathan & Guthrie, 1998a)+ Western blotting showed that the epitopetagged Prp24(HA) 3 protein was reproducibly immunodepleted in three trials to give postdepletion levels of Prp24(HA) 3 protein that were 15 25% of mock- and pre-depletion levels (data not shown)+ After depletion or mock depletion of Prp24 protein, all samples were treated with [a- 32 P]-body-labeled wild-type U6 RNA to visualize the newly generated U6 snrnp complexes+ Some U6-reconstituted samples were further treated with recombinant Prp24 protein+ The 32 P-labeled U6 snrnp-containing complexes were separated on a native gel (Raghunathan & Guthrie, 1998a), and newly formed U4-U6 U5 tri-snrnp complex was clearly discernable as the slowest migrating and most intense complex on the gel and the only complex shifted (to slower mobility) by the addition of cold actin pre-mrna (Fig+ 6A), as observed by Raghunathan and Guthrie (1998a), although they used northern probes rather than 32 P-labeled U6 to visualize U6 snrnp complexes+ In our experiments, quantitation of de novo tri-snrnp complexes revealed that the assembly of 32 P-labeled U6 RNA into U4-U6 U5 tri-snrnp complexes was catalyzed by Prp24 protein+ For extracts depleted of both endogenous U6 RNA and Prp24 protein in three trials, addition of 32 P-labeled U6 RNA but not Prp24 protein produced only 14 16% as much de novo U4-U6 U5 tri-snrnp as produced in samples that were mock depleted of Prp24 but otherwise treated equivalently (Fig+ 6A, cf+ lanes 3 and 5)+ In split samples of the U6-, Prp24-depleted extracts, addition of both 32 P-labeled U6 RNA and recombinant Prp24 protein restored assembly of new U4-U6 U5 tri-snrnps to levels that were 75 97% as high as for samples that were mock depleted of Prp24 but otherwise treated equivalently in three trials (Fig+ 6A, cf+ lanes 4 and 6)+ We conclude that Prp24 protein is important or required for catalyzing the assembly of naked U6 snrna into U4-U6 U5 tri-snrnp in our extracts, presumably by binding to U6 snrna to facilitate its incorporation into U4-U6 disnrnp as demonstrated previously in vitro (Raghunathan & Guthrie, 1998a)+ Mutations in the telestem sequences or in the 39 terminal uridine tract of yeast U6 snrna affect U6 RNA protein binding and U4-U6 U5 tri-snrnp assembly Having verified that Prp24 protein catalyzes the assembly of naked U6 RNA into U4-U6 U5 tri-snrnps in vitro, we were interested in determining whether our U6 telestem and 39 terminal domain mutations affect U6 binding of Prp24 protein and/or Lsm proteins and also whether tri-snrnp assembly is affected+ Mass spectrometry analysis of purified yeast U6 snrnp indicated that free U6 snrnp has a relatively small complement

11 59 and 39 domains of yeast U6 snrna 1021 FIGURE 6. A: U4-U6 U5 snrnp assembly for U6-, Prp24-depleted yeast extract reconstituted with wild-type 32 P-labeled U6 RNA and treated or not with recombinant Prp24 protein+ Control samples were mock depleted of endogenous Prp24 protein (lanes 2 4)+ Yeast splicing extract was treated with d1 oligonucleotide to digest endogenous U6 snrna as described in Figure 2+ Aliquots of the U6-depleted extract were incubated (lanes 5 6) or not (lanes 2 4) with 12CA5 antibody to endogenous epitope-tagged Prp24(HA) 3 protein in the extract+ Aliquots of the mock- or Prp24-depleted, U6-depleted extract were treated with 32 P-labeled U6 RNA and assayed as in Figure 4B+ For lane 2, actin pre-mrna was added just prior to the 20-min incubation+ B: U4-U6 U5 snrnp assembly for U6-depleted yeast extract reconstituted with mutant and wild-type 32 P-labeled U6 RNAs+ U6 mutations were introduced in the telestem sequences (nt and 86 95) of yeast U6 RNA+ Samples were prepared as in Figure 4B+ In synthetic U6 RNAs marked with an, the last four linkages at the 39 ends were racemic phosphorothioates+ of stably associated proteins comprising only Prp24 and the Lsm2 8 protein complex (Stevens et al+, 2001)+ To assay binding of U6 snrnp proteins to our mutant U6 RNAs, we prepared yeast extracts that included an epitope tag on the sole copy of the Prp24 protein or on the sole copy of the Lsm4 protein (of the Lsm2 8 complex)+ These epitope-tagged extracts were reconstituted with a variety of individual U6 RNAs mutated in the 39 terminal domain, the 39 stem-loop, and/or the telestem sequences of U6 as listed in Table 3+ Binding of the mutant U6 RNAs, carrying 32 P-radiolabels, to epitope-tagged Prp24 or Lsm4 protein was assayed by coimmunoprecipitation and scintillation counting of the bound U6 RNAs+ This was followed by denaturing gel electrophoresis of the coimmunoprecipitated U6 RNAs and their supernatants to monitor and ensure the stability of U6 RNA in each sample (data not shown)+ The same 32 P-labeled U6 RNAs were also tested for their assembly into U6-containing snrnps in vitro, including U4-U6 U5 tri-snrnps, by using Raghunathan and Guthrie s (1998a) native gel electrophoresis system+ Correlation of U6 RNA protein binding affinity, U6 snrnp assembly profiles, and splicing activity for each mutation provides a more comprehensive understanding of the various roles of the 39 domain, the 39 stem-loop, and the telestem sequences of U6 snrna+ The most obvious effect of mutating the upstream or downstream telestem sequence is that, under the immunoprecipitation conditions (150 mm NaCl), binding of the telestem mutant U6 RNAs to Prp24(HA) 3 was substantially diminished relative to such binding for wildtype U6 (Table 3)+ For the downstream telestem mutant, U6+39C,86 95polyC, this mutational disruption of the telestem also diminished Lsm protein binding, trisnrnp assembly and splicing, all to 40 60% of wildtype U6 levels (Table 3; Fig+ 3B, lanes 7 and 12; Fig+ 6B, lane 4)+ In comparison to the downstream mutant, the upstream telestem mutant, U polyC,86C, under parallel conditions showed ;1+5-fold less binding of Lsm proteins and ;1+6-fold less splicing activity than the downstream mutant, as well as a notable 3+5-fold reduction in the amount of tri-snrnp produced (Table 3; Fig+ 6B, lane 3)+ For either upstream or downstream telestem mutants, the striking reduction in their binding of Prp24 protein (relative to wild-type U6) is consistent with the U6 snrnp assembly defects observed for these mutants by native gel electrophoresis: The telestem disruption mutations caused an accumulation of naked, mutant U6 RNA (Fig+ 6B, lanes 3 and 4 as observed on a lighter exposure) that comigrated with synthetic U6 RNA alone on the same gel (not shown)+ The telestem mutations also caused an accumulation of free U6 snrnp complexes, including a complex that had unusually fast mobility on the gel (Fig+ 6B, lanes 3 and 4) compared to mutant U6 snrnps with a U80G or G81C mutation (cf+ Fig+ 6 in Ryan & Abelson, 2002)+ This accumulation of naked, mutant U6 RNA and mutant U6 snrnps is most likely a result of diminished binding affinity for Prp24 protein that is necessary for U6 snrnp formation (Stevens et al+, 2001) and for catalyzing U4-U6 snrnp assembly (see above; Raghunathan & Guthrie, 1998a)+ Also, the unusually fast mobility of some of the mutant telestem U6 snrnps may reflect their incorporation of Lsm4 protein without also incorporating Prp24 protein, as suggested by the coimmunoprecipitation results for these same U6 telestem mutants (Table 3)+ To further understand the effects of the 39 terminal uridine tract mutations, we analyzed the abilities of these RNAs to form U6 snrnp, U4-U6, and U4-U6 U5 snrnps by native polyacrylamide gel electrophoresis+ We also tested their binding of Lsm core proteins as well as of Prp24 protein via coimmunoprecipitation experiments+ Although we had found that mutation of the four U6 terminal uridines to polyadenine or polycytidine (U polyA/polyC) had only a minor effect on

12 1022 D.E. Ryan et al. splicing yields, these mutations had a substantial effect (10-fold) on the binding of Lsm proteins (Table 3)+ This binding effect was expected, as studies of human U6 had shown that its terminal uridine tract was critical for binding to human Lsm proteins (Achsel et al+, 1999)+ Surprisingly, our U6 uridine tract mutants under our same coimmunoprecipitation conditions were similarly inhibited for binding to Prp24 protein+ This finding provides strong evidence for the suggestion from UV crosslinking and yeast two-hybrid studies that U6 binds Prp24 and Lsm proteins cooperatively (Vidal et al+, 1999; Fromont-Racine et al+, 2000)+ Furthermore, we found that U6 binding of Prp24 depends on or is strongly enhanced by U6 binding of Lsm proteins+ Despite the reductions in binding of Lsm and Prp24 proteins to U6 RNA with 39 uridine tract mutations, only minor effects were observed for U4-U6 U5 tri-snrnp assembly (Fig+ 5B, lanes 4 and 5; Table 3), consistent with the minor effects on splicing found+ Examination of the native gel showed some accumulation of the naked mutant U6 RNAs in the reconstituted samples relative to the wild-type U6 control (Fig+ 5B, lanes 4 and 5 vs+ lane 3)+ When the remainder of the U6 39 domain was mutated along with the 39 terminal uridine tract (U polyC and U polyC polyA), U6 binding of Prp24 and Lsm4 proteins was diminished by 10- and 50-fold relative to wild-type U6 for each protein, respectively, and U4-U6 U5 tri-snrnp assembly and splicing were severely inhibited (Table 3; Fig+ 4A, lanes 12 and 13; Fig+ 5B, lanes 7 and 8)+ As expected for loss of Lsm and Prp24 protein binding to these 39 domain mutants, the native gel showed substantial accumulations of naked mutant U6 RNAs not incorporated into U6 snrnps (Fig+ 5B, lanes 7 and 8)+ In contrast, polycytidine mutation of the entire 39 domain except for the terminal uridine tract (U polyC) showed a 12-fold higher normalized level of Lsm protein binding relative to that for mutation of the entire 39 domain (U polyC)+ This U polyC mutant also showed a 2+5-fold higher normalized level of Lsm protein binding relative to that for mutation of only the terminal uridine tract (U polyC or U polyA) its mutational complement for the 39 domain (Table 3)+ These results confirm that the 39 terminal uridine tract is a major determinant for binding to Lsm proteins in yeast U6+ In contrast to the 2+5-fold higher normalized level observed for binding of Lsm proteins to U polyC, binding of Prp24 protein to this mutant did not increase in a parallel fashion, presumably because mutation of U6 nt includes mutation of nt 86 95, the 39 sequence of the telestem, a sequence that is important for normal Prp24 binding (see above)+ This hindrance of Prp24 binding is consistent with an accumulation of naked mutant U6 RNA that was not incorporated into U6 snrnps (Fig+ 5B, lane 6)+ Our data suggest that although U6 binding of Prp24 protein depends on or is strongly enhanced by U6 binding of Lsm proteins, the binding of Lsm proteins is not dependent on U6 binding to Prp24+ This point is most dramatically illustrated by the telestem disruption mutations, U polyC,86C and U6+39C, 86 95polyC, which showed a substantial fold reduction in binding of Prp24 protein relative to wildtype U6 RNA+ In contrast, under the same conditions in vitro, these two telestem mutations showed a fairly modest 2 3-fold reduction in binding of Lsm proteins relative to wild-type U6 RNA (Table 3), demonstrating that Lsm protein binding is not dependent on U6 binding to Prp24 protein+ Compensatory mutations in the U6 telestem affect U6 RNA protein binding, U4-U6 U5 tri-snrnp assembly, and pre-mrna splicing in vitro Our results demonstrated that mutational disruption of the entire U6 telestem blocks the normal binding of U6 RNA to Prp24 protein, whereas complete mutation of the upstream strand of the telestem also inhibits assembly of U4-U6 U5 tri-snrnp (Table 3)+ Despite these notable effects on Prp24 function, mutational disruption of the entire telestem can have a rather modest effect on splicing of two- to threefold (Table 3)+ If disruption of the telestem is a necessary event for spliceosome activation, then telestem disruption mutations are perhaps unlikely to produce substantial splicing defects+ On the other hand, mutations that hyperstabilize the telestem helix might block spliceosome assembly and/or splicing and provide evidence for the presence and function of the U6 telestem in splicing, as has been observed for hyperstabilizing mutations in the U6 39 stem-loop (Madhani et al+, 1990; Wolff & Bindereif, 1993, 1995; Fortner et al+, 1994)+ To test hyperstabilization of the telestem in vitro, we prepared U6 RNAs in which three contiguous A-U base pairs in the telestem were mutated to A-C mismatches or G-U wobble pairs or were fully transverted to G-C base pairs+ These telestem mutants were U polyC, U polyG, U polyC, U polyG, and two potentially hyperstabilizing combinations, U polyC,93 95polyG, and U polyG,87 89polyC+ In tandem experiments, we prepared a set of U6 telestem mutations that could restore telestem base pairing without hyperstabilization+ In this set of constructs, three contiguous A-U base pairs in the telestem were mutated to A-A or U-U mismatches or were fully inverted to U-A base pairs+ These telestem mutants were U polyA, U polyU, U polyA, U polyU and two potentially compensatory combinations, U polyA,93 95polyU, and U polyU,87 89polyA+ The transverted and inverted telestem mutant U6 RNAs were assayed in U6-depleted extract for their